Switcher By-Pass

[itsmoked]

Greetings. Here's a problem I've been working on for a few days. I have a nice little switcher that can convert anything between about 14V and 75V to 12VDC to run some lousy fans and a processor that has sub-regulation from the 12V. This means the 12V can be anywhere from say 11V to 14V and nothing would care in the least i.e. lousy regulation is fine.

The catch.. There's always a catch. I want this product to be able to run from 12V as well as 15,24,36,48V, etc.

Considered solutions. Run a buck/boost instead of a buck. Unfortunately that results in a switcher twice as big and 7 dollars instead of the above one which is only about 4 dollars.

Do you think something like the little network below the switcher with the two transistors will get the job done by essentially bypassing the switcher whenever the voltage is below 14V.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

There could be a problem with Q6 running as an emitter follower. You need some current through R5 - which means that you drop some voltage there (quite a lot if you want to keep dissipation in R5 low). Then the Q6 base will perhaps be a volt below input voltage. If you need some current out of the Q6 emitter, you will lose another volt between base and emitter. So, there's a risk that it won't work with 12 V input if you need an 11 V or even 10 V output.

If you can use a PNP for Q6 and change the control part accordingly, I think the chances are better. The U_ce can then be as low as a few hundred millivolts.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[MacGyverS2000]

SEPIC, maybe?

Dan - Owner

http://www.Hi-TecDesigns.com

 

[OperaHouse]

And just why won't it run below 14V, does the chip have a low voltage shutdown? You didn't mention what the chip was. There are ways to fool some of this chips, the TNY series of line operated chips can be easily fooled to work on 12V. I saw a car voltage booster that put several diodes in series to drop the voltage just so the boost regulator would still work.

 

[itsmoked]

There could be a problem with Q6 running as an emitter follower. You need some current through R5 - which means that you drop some voltage there (quite a lot if you want to keep dissipation in R5 low). Then the Q6 base will perhaps be a volt below input voltage. If you need some current out of the Q6 emitter, you will lose another volt between base and emitter. So, there's a risk that it won't work with 12 V input if you need an 11 V or even 10 V output.

If you can use a PNP for Q6 and change the control part accordingly, I think the chances are better. The Uce can then be as low as a few hundred millivolts.

Thanks for pointing out that detail Skogs. Got me thinking the whole bypass trip point might be too flaky so I've re-tooled it using a comparator with a single chip milliamp switcher to run the comparator and set the reference. Doing this also solves another problem which is thermally controlling one of the fans (for free! Spare comparator.). This change adds about $1.50 to the cost which is better than $3+.

SEPIC, maybe?

That would probably work too but cost me days of mucking about and prototyping.

And just why won't it run below 14V, does the chip have a low voltage shutdown? You didn't mention what the chip was. There are ways to fool some of this chips, the TNY series of line operated chips can be easily fooled to work on 12V. I saw a car voltage booster that put several diodes in series to drop the voltage just so the boost regulator would still work.

Hi Opera. The chip is the LM5010.
It doesn't actually say it won't work. It doesn't do its own UVLO until down about 5.25V I wish it would just turn on its switch and sit there feeding-thru whatever is available but I think it will interject x us of off period causing some sort of negative issue. The plan is to add this other circuitry then crank a board and try it. If it does work adequately I'll not be stuffing the Bypass.



Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

Have you seen the T.I. Webbench webpage for the LM5010? I've not gone into it beyond the front page, but it's promising a reference design to meet your requirements for wide input voltages. I've dialed in 10 to 75 Vin and 12 Vout, and it's promising a BOM of $2.64, etc.

Apologies in advance if you're way past that point.

 

[itsmoked]

Hi VE1BLL!

Workbench was invented by National and I'm glad TI hasn't screwed it up yet - like everything else they suck up - but I digress. That's exactly where this design came from. There is one single design that encompasses the 12V in 12V out regulated requirement. That one results in the $7++ cost.

Their BOM estimates are off by 50%. Add 50% to them and they're pretty good.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

"...encompasses the 12V in 12V out..."

The GUI that I was looking at allowed setting Vin to a range as wide as 8v - 75v (not just "12V in"). Given that the issue is the Vin range below 12v, does this minor discrepancy in the statement of Vin requirements lead anywhere useful?

Note: I'm assuming that there are more than one reference design architecture, depending on how you set the requirements fields. That's the impression given to me by the webpage, but I've not gone in deeper to confirm.

Last resort - I've never met an Application Engineer (at the component OEM, such as TI) that I didn't like. They've always been perfectly helpful. Does the modern TI still offer such phone-in technical support? If one of their reference designs is supposed to work down to 8V input, and it doesn't, then this is where they would usually be willing to help review things.

Observation: The Vf voltage drop on D5 is pure waste. Is D5 a low voltage drop Schottky diode? Such a diode might gain you about one-half volt in the difference.

 

[iop995]

Depending of CPU current and capacitors bank that supply CPU step-down converter, may use a relay with 2-3 series diodes and a comparator that swith on relay and switch off LM (shutdown pin low) when Vin is below 14V. Instead of relay may be used a SCR also, but add some complication to turn-off it. For relay variant maybe need to add more capacitors for CPU converter to avoid reset.

 

[itsmoked]

"...encompasses the 12V in 12V out..."

The GUI that I was looking at allowed setting Vin to a range as wide as 8v - 75v (not just "12V in"). Given that the issue is the Vin range below 12v, does this minor discrepancy in the statement of Vin requirements lead anywhere useful?

Note: I'm assuming that there are more than one reference design architecture, depending on how you set the requirements fields. That's the impression given to me by the webpage, but I've not gone in deeper to confirm.

Last resort - I've never met an Application Engineer (at the component OEM, such as TI) that I didn't like. They've always been perfectly helpful. Does the modern TI still offer such phone-in technical support? If one of their reference designs is supposed to work down to 8V input, and it doesn't, then this is where they would usually be willing to help review things.

I'm sorry if I wasn't clear that this product needs to run on anything between 12V and 50V so 8V is completely useless to me. The 12V is the whole problem since I need 12V out the other side. The one single design in workbench that can provide that functionality is the LM5018 which costs considerably more and requires a double fist-full of surrounding parts to work. If lower voltages like 7 or 8V needed to be supported it would be a shoe-in but just to allow by-passless operation it's overkill.

Observation: The Vf voltage drop on D5 is pure waste. Is D5 a low voltage drop Schottky diode? Such a diode might gain you about one-half volt in the difference.

"Pure waste" unless you provide a warranty in which case it's pure gold.

And yes, it is an (expensive) Schottky to ease your mind. LOL

Keith Cress
kcress -

http://www.flaminsystems.com

 

[itsmoked]

Depending of CPU current and capacitors bank that supply CPU step-down converter, may use a relay with 2-3 series diodes and a comparator that swith on relay and switch off LM (shutdown pin low) when Vin is below 14V. Instead of relay may be used a SCR also, but add some complication to turn-off it. For relay variant maybe need to add more capacitors for CPU converter to avoid reset.

Thanks iop. A relay would definitely cost more than a 30 cent FET. So I've opted for the FET. This think is only needing about 500mA.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[iop995]

Sure, FET is much better.

 

[VEBill]

[link itsmoked]...needs to run on anything between 12V and 50V so 8V is completely useless to me. ...[/url]

I didn't write " 8V "; I wrote "...a range as wide as 8v - 75v...". That 8v-to-75v range more than encompasses "anything between 12V and 50V". Am I misunderstanding something? It seems clear enough.

Here's the reference schematic diagram that was offered up on the LM5010 page. The diagram indicates that the circuit should accept any Vin from 8 volts all the way up to 75 volts. Their offered information is consistent.

If I were you, I'd want to know why the circuit as you've designed and built isn't working down to about 8 volts. It's an interesting discrepancy, because if it leads to a solution (such that your regulator worked over the range of Vin), then you wouldn't need the bypass. Right?

PS: I'm trying to be helpful. If this discrepancy isn't helpful for some reason that I'm missing (miscommunication?), then apologies in advance.

 

[VEBill]

Observation: Later in the data sheet (

http://www.ti.com/lit/ds/symlink/lm5010.pdf),

the 'Final Circuit' is marked "15 - 75V Input". I don't see where they explain how to achieve the lower end (8V-...) of the supposed 8-75V Vin range.

 

[ScottyUK]

VE1BLL,

Might be wrong but as this is a step down[/] regulator then the output is by default lower than the input. If you want 12V out you have to feed it something higher than 12V. The converter doesn't have boost capability, it's a fairly straightforward buck converter.

 

[IRstuff]

The title of the 5010 datasheet contains "Step-Down," which can also be discerned from the block diagram on page 5. The driver for the output voltage is an n-channel device running as a source follower, tied directly to Vin, so the best case output voltage is Vin-Vth, which means that at Vin=12V, the output is probably around 10.8V, based on the 1.2V UVLO, which looks to be the threshold of the n-channel devices on the chip.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[VEBill]

Okay. Sorry for the confusion.

 

[OperaHouse]

I don't think that is correct. The high side drive is a bootstrap so the drive is always about 7ish from internal regulator and below 8V can turn off. Pin SW has to go near to 0V in order to charge C4, the boot cap. If the internals never allow 100% dury cycle, that cap will still charge and there will be full drive. Below the set voltage it will just follow the input voltage with some IR loss till it reaches turnoff at about 8V.

 

[VEBill]

DX.com 126106

$3.80 shipped

Same website has quite a few similar products.

 

[itsmoked]

I don't think that is correct. The high side drive is a bootstrap so the drive is always about 7ish from internal regulator and below 8V can turn off. Pin SW has to go near to 0V in order to charge C4, the boot cap. If the internals never allow 100% dury cycle, that cap will still charge and there will be full drive. Below the set voltage it will just follow the input voltage with some IR loss till it reaches turnoff at about 8V.

Well that seems to make some sense. I'll let you know as I just sent that design out.




Keith Cress
kcress -

http://www.flaminsystems.com